Long-lived identity hijacking through stolen refresh tokens—one of the stealthiest cloud persistence techniques threatening modern enterprises.
Understanding the Threat
This breach pattern exploits long-lived refresh tokens to maintain persistent, stealthy access to cloud and SaaS environments for extended periods—sometimes months. Unlike traditional session hijacking, refresh token persistence operates beneath detection thresholds, silently regenerating valid credentials.
Attackers acquire these tokens through infostealers, endpoint compromise, M365 extraction, OAuth theft, or browser cookie harvesting. Once obtained, they weaponize them for indefinite impersonation.
Why Refresh Tokens Are Dangerous
Bypass multi-factor authentication completely
Persist through password resets
Work seamlessly across devices
Operate for 30–90 days by default
Survive browser restarts and reboots
Rarely monitored by security teams
Enable silent, on-demand token generation
Attack Mechanics & Persistence
1
Initial Compromise
Attackers deploy infostealers, exploit endpoint vulnerabilities, or harvest tokens from browser storage to obtain valid refresh tokens.
2
Token Extraction
Refresh tokens extracted from M365 sessions, OAuth flows, federated authentications, or developer workstations with cloud CLI tools.
Identify identical tokens replayed from geographically dispersed or anomalous IP addresses within suspicious timeframes.
02
DL-041: Lateral API Token Issuance
Detect stolen refresh tokens rapidly generating multiple access tokens across different services and APIs.
03
DL-025: Impossible Travel Patterns
Flag access tokens created from distant geographic locations within physically impossible time windows.
04
DL-099: Session Extension Anomalies
Monitor privileged sessions silently extended far beyond typical user behavior or policy thresholds.
Detection Challenges
Refresh token abuse generates minimal authentication logs and often appears as legitimate user activity, making behavioral analytics and token lineage tracking essential for detection.
Attack Chain Integration
BP-041 spans multiple stages of the Identity Attack Chain, forming the foundation for long-term cloud compromise.
1
Stage 3
Credential Acquisition
Initial token theft via infostealers or endpoint compromise
2
Stage 6
Token Tampering
Session hijacking through refresh token replay
3
Stage 8
Identity Persistence
Establishing durable access through token regeneration
4
Stage 9
Action on Objectives
Data exfiltration, privilege escalation, lateral movement